跳到正文
Hacker News:AI 熱帖·· 12 小時前AI 評分36

Together.ai:請求提示詞洩漏

Together.ai: Prompt Leakage for Requests

AI 導讀

Together.ai 的提示詞洩漏事件引發討論:一名評論者指出,開發公開 GitHub 儲存庫 xorl 時,AI coding agent 提交了可存取私人 Hugging Face 儲存庫的憑證。評論者亦稱,相關提示詞雖涉及 Zero Data Retention,仍被保留、分享予其他公司並由人員閲讀;有人建議敏感工作可考慮自行託管模型。

正文

Image 1Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Together.ai: Prompt Leakage for Requests (together.ai) 19 points by Fe2O34 hours ago | hide | past | favorite | 7 comments help Image 2tczMUFlmoNk2 hours ago | next[–] > During development of our public GitHub repository, xorl, an AI coding agent committed a credential that provided access to the private Hugging Face repository My emphasis on the accountability laundering. No, "xorl" did not do this—you did. And, if "xorl" is https://github.com/togethercomputer/xorl and togethercomputer is together.ai, then you wrote "xorl" and are doubly responsible. If five years ago I had said, "During development of our public GitHub repository, my computer decided to commit a credential", this would have been immediately recognized for the embarrassing abdication of duty that it is. So, too, should this be now. reply Image 3ashishb1 hour ago | parent | next[–] How come they are not using any secret leak prevention mechanisms in their pre-commit hooks? https://ashishb.net/programming/python-in-production-2/#prev... reply Image 4sans_souse2 hours ago | prev | next[–] Am I inferring correctly here, in that Dragos basically had some sort of experiment set up and running for exactly this scenario, that effectively allowed them to trace the breach to the individual? reply Image 5jerrythegerbil2 hours ago | parent | next[–] You are inferring correctly. https://www.dragos.com/blog/ai-supply-chain-security-canary-... reply Image 6pyaamb1 hour ago | root | parent | next[–] But their model wasn't reply Image 7nsagent1 hour ago | root | parent | prev | next[–] That's a truly ingenious and scary investigation. Zero Data Retention is supposed to mean something. In practice, these prompts were retained, shared with other companies, and read by people. The first bit of advice is on point: Host it yourself, if you can afford it. This is the only option where nobody else gets to read your prompts, but the hardware isn’t cheap. A Mac Studio with 512 GB of memory starts at $22,299, NVIDIA’s DGX Spark now costs $6,950, a DGX Station starts at about $105,000, and an eight-GPU B300 server lists for $570,000 or more. For most teams, this only makes sense for the most sensitive work. Despite being an NLP researcher, I never used LLMs for coding until this summer. Glad I exclusively used locally hosted models. My M5 Ultra 512GB just arrived and I'm hoping it turns out to be a wise decision for my startup. reply Image 8nickandbro2 hours ago | prev[–] The mass hacking continues… reply Image 9 Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact Search:

來源:Hacker News:AI 熱帖 · news.ycombinator.com